With a nofailed Stage now not writing a checkpoint (correct, IMO), Stages downstream will be run even if they've already been done. Typically, I'll be marking a stage as nofail only if it has no downstream dependent tasks, so in general I don't want these rerun if I just rerun a nofailed stage to "fill in" the pipeline's steps.
With a
nofail
edStage
now not writing a checkpoint (correct, IMO),Stage
s downstream will be run even if they've already been done. Typically, I'll be marking a stage asnofail
only if it has no downstream dependent tasks, so in general I don't want these rerun if I just rerun anofail
ed stage to "fill in" the pipeline's steps.